TapCub is live — analytics, insights and live chat, free on one platform
Glossary

What is cross-device tracking? How logins connect devices

Cross-device tracking links one person's activity across phone, laptop and tablet. Deterministic vs probabilistic linking, a worked example of how many visitors you overcount without it, and how login-based linking behaves.

Emma Collins Reviewed by Chen Yu Also called cross-device measurement Last updated Nov 17, 2026 Glossary

Quick answer

Cross-device tracking is linking one person's activity across several devices, such as a phone, a laptop and a tablet, so it is analyzed as one person. Deterministic linking uses something the person supplied, usually a login. Without a link, each device counts as a separate visitor.

Formula: Overcount = devices ÷ people − 1. In the example, 1,340 ÷ 1,000 − 1 = 34%.

What is cross-device tracking?

Cross-device tracking is linking the activity of one person on several devices, such as a phone, a laptop and a tablet, so that it is analyzed as one person rather than several. The US Federal Trade Commission describes it as associating multiple devices with the same consumer. Without it, one shopper who browses on a phone and buys on a laptop is counted as two visitors, and the purchase looks as if it came from nowhere.

There are two ways to make the link, and the difference matters more than any other point on this page. Deterministic linking uses something the person supplied, usually a login: when the same account signs in on both devices, they belong together. Probabilistic linking guesses from signals such as network address or location. This page is about the first, because it is the one you can explain to a user, test, and fix when it goes wrong.

Deterministic and probabilistic linking

Deterministic (login-based)Probabilistic (inferred)
What links the devicesThe same account identifier on eachA statistical guess from shared signals
AccuracyHigh for the people who log inVaries; false matches are possible
CoverageOnly people who sign inEveryone, including people who never do
Transparency to the userThe person knowingly signs inOften happens without the person noticing
Failure modePeople who do not log in stay separateTwo people on one network can be merged

The FTC staff report on this topic draws exactly this line, and recommends that companies be transparent about the practice and offer meaningful choices. Google Analytics is an example of the deterministic route: its User-ID feature lets you attach your own identifier to a user so activity is connected across sessions and devices, and the identifier must not contain information a third party could use to identify the person.

Why you need it: a worked example

Sample data: 1,000 shoppers visit a store over a month. 660 use one device. 340 use two (a phone and a laptop). So the number of devices visiting is 660 + 2 × 340 = 1,340.

  • A tool that counts devices reports 1,340 visitors. That is 34% more than the 1,000 real people.
  • Of the 340 two-device shoppers, 220 are logged in on both devices. Linking those devices removes one visitor each, leaving 1,340 − 220 = 1,120 people. The other 120 two-device shoppers never logged in on both, so they still appear twice. The overcount falls from 34% to 12%.
  • 60 purchases happened in the month. Divided by devices, the conversion rate is 60 ÷ 1,340 = 4.5%. Divided by linked people it is 60 ÷ 1,120 = 5.4%. Divided by real people it would be 60 ÷ 1,000 = 6.0%.

Two lessons. First, every metric that divides by people (conversion rate, revenue per visitor, retention) is understated when people are counted per device. Second, the fix is partial: login only joins the people who log in. The remaining gap is not a bug; it is the part nobody can see without guessing.

What login actually links

The behavior of a login-based link is worth knowing in advance.

  • The first login on a device backfills that device's earlier anonymous history onto the person. The funnel from first visit to sign-up then becomes one path.
  • A second account on the same device does not take over old data. Rows that already belong to a person stay with that person; the device then points at the new person and only new events go to them.
  • A shared computer does not leak history. Signing out clears the login and starts a clean session.
  • Placeholder IDs are rejected. Values such as guest, anonymous, null, all zeros and -1 would merge unrelated people into one, so they are refused.
  • Devices per person have a limit. The default is 50; beyond that, the oldest is unbound.

Common mistakes

  • Expecting logged-out visitors to link. Visitors who never log in stay separate in the aggregates. That is a limit of the method.
  • Calling the login before sign-in succeeds. The ID should be set after your own authentication works, not when the form is shown.
  • Using one ID for many people. A shared account or a placeholder merges strangers, and the history of one becomes the history of all.
  • Using different ID types on each platform. Web, app and mini program only merge when they send the same type and the same string; otherwise they are three people.
  • Sending personal data as the ID. Use an internal ID. If you use an email as the identifier, hash it (TapCub does this for you), and disclose identifier use in your privacy policy.
  • Treating the merged count as the only truth. The drop in unique visitors after linking is a different way of counting, not lost traffic.
  • Ignoring consent. Login linking builds on a device ID, and TapCub writes that only when the consent and privacy setup allow it.

For the techniques that join identifiers, see identity resolution and identity graph. The call that reports who someone is has its own page, identify, and the two identifiers being joined are covered in user ID and device ID.

Cross-device tracking in TapCub

Take the 1,000 shoppers. This is how to get from 1,340 visitors to 1,120 people.

  1. Check the mode. Linking by login only works when the project is in identified mode. Configuration › Privacy center shows whether the project is anonymous or identified, together with consent and the collection tier. Whether a project can link people this way depends on the settings in the Privacy center. In anonymous mode, nobody is profiled, and visits are still counted.
  2. Report the login. After your own sign-in succeeds, call tapcub.login('u_123', { $name: 'Ada' }); on sign-out, call tapcub.logout(). The ID can be up to 200 characters; emails are lowercased and hashed, and phone numbers lose spaces and dashes. bind() adds more identifiers to the same person. Details: call login and bind.
  3. Use the same ID everywhere. A website, an app and a mini program merge into one person when they send the same type and string. On a mini program, call login with the OpenID type, and bind a UnionID to link the platforms.
  4. See one row per person. In All-platform analytics › Users › People, open the profile: identity links, devices, and a timeline with a row for each day that merges the bound devices. Activity from before identification appears as an anonymous segment.
  5. Check how much it removed. All-platform analytics › Users › Identity map has a difference card comparing visitors, devices and people for the dates you pick, including how many extra counts came from one person on several devices. The mapping log records every bind, split, rejection and backfill, and an owner or admin can split a device apart. Merge rules are in how one person is merged across devices.
  6. Switch the counting mode. Counting modes let a report count visitors, devices or users. Switching from visitor to user mode merges one person's phone and laptop, so the number usually drops: a different count, not missing data.

People who never log in stay in the aggregates and get no profile. TapCub does not link them by guessing.

Sources

Was this explanation clear?

We use your answers to fix and expand this page.

Emma Collins

Product Analyst

Focuses on event analysis, conversion funnels, and retention, using clear metric definitions and structured analysis to explain product performance.

FAQ

Questions people ask

Short answers to the follow-up questions that come up most.

Still have a question?

Chat with the team behind TapCub — we usually reply within a few hours on business days.

What is cross-device tracking?

Cross-device tracking is linking one person's activity across devices such as a phone, a laptop and a tablet so it is analyzed as one person. Without it, a shopper who browses on a phone and buys on a laptop is counted as two visitors.

What is the difference between deterministic and probabilistic cross-device tracking?

Deterministic tracking links devices through something the person supplied, usually a login. Probabilistic tracking infers links from signals such as network address or location, so it covers more people but can match the wrong ones.

Can you track a user across devices without a login?

Not reliably. Without a shared identifier, devices stay separate visitors, and any link would be a guess. Login-based linking joins only the people who sign in; the rest remain separate in the totals.

Why did my visitor count drop after I turned on identity?

Linking a person's phone and laptop merges two visitors into one, so unique counts usually fall. That is a different way of counting, not lost traffic, and rates such as conversion per person usually rise.

See this number on your own site

Add one script tag for web and apps, and the first reports fill in within minutes.

See your data clearly. Find your growth.

Every click, backed by data. Install one line of code and see your first numbers in a minute.

No credit card · Free plans for analytics and chat · Cookieless analytics